Project Report to Defra: Wheat Breeding AR 0914

Project report to Defra: Wheat Breeding AR 0914

Introduction

Traditional plant breeding programmes rely on selecting genotypes in optimal conditions: in the absence of weeds, diseases and pests and with peak nutrient availability. This breeding approach has produced many successful pedigree line bred varieties for non-organic production systems. However, organic agriculture has suffered from a lack of varieties adapted to the environmental variability on organic farms (Wolfe et al., 2008). Furthermore, there is a need to reduce costly inputs in non-organic agriculture and to prepare for the impacts of climate change.

In organic and low input conditions, the variability of the environment has a far greater influence on yield than the choice of variety (Wolfe et al., 2008). This can lead to a lack of stability in performance that has been demonstrated in many studies (e.g. Soliman & Allard, 1991). Physical mixtures of complementary varieties provide an improved ability of a crop to buffer variation in soil, climate and disease and weed pressures (Wolfe, 2001, Didon & Rodriguez, 2006).

The advantage of diversity in cereal variety mixtures has been well demonstrated (Finckh et al., 2000, Finckh and Wolfe, 2006). However, there are practical limits to the number of genotypes that can be used, usually no more than three or four, which limits the potential for buffering against environmental variability.

This project was designed to overcome the problem of limited genetic variability by developing crop material directly from segregating populations of half-diallel intercrosses involving up to 20 parents in the form of 'evolutionary breeding' (Suneson, 1956).

Objectives

The overall objective of this project was to increase the sustainability and competitiveness of both non-organic and organic farming systems by developing genetically diverse wheat populations that respond rapidly to on-farm selection for improved productivity and yield.

The five main objectives, as stated in the contract, were:

1. To generate six distinct, highly heterogeneous composite cross populations of winter wheat for further development and selection. The populations will comprise: one with parental material selected for good milling potential, one with parents selected for high yield potential and one comprising both sets of parent material. Each of these populations will then be split to either include or exclude heritable male sterility.
2. To evaluate the performance and evolution of composite cross populations over time under a diverse range of environmental conditions and identify characteristics that confer improved productivity in these environments.
3. To track the genetic changes that accompany selection, so providing a better understanding of the assemblages of traits that underlie improved productivity in diverse environments.
4. To provide genetically diverse crop material for further selection by farmers and as a resource for future publicly funded research.
5. To disseminate the results to the scientific community and industry.

Objectives One, Two, Four and Five have been met fully. Objective Three is ongoing.

Methods

Selection of parents

Six composite cross populations of winter wheat were created in 2002 by John Innes Centre (JIC). The twenty parental cultivars were selected from data of both published and unpublished studies, and from experience of the consortium partners. Key criteria for selection included a diverse genetic base and thepotential forrobust performance under low input agronomic conditions. The parent cultivars were selected in summer 2002 in two categories: high yielding (Bezostaya, Buchan, Claire, Deben, High Tillering Line, Norman, Option, Tanker, Wembley) ‘Yield’ varieties, and high bread making quality (Bezostaya, Cadenza, Hereward, Maris Widgeon, Mercia, Monopol, Pastiche, Renan, Renesansa, Soissons, Spark, Thatcher) ‘Quality’ varieties. Bezostaya was included in both categories as it was known as both high yielding and high quality in Russia, where it was grown successfully over many years.

Creation of CCPs

All 20 parents were crossed together in a complete half-diallel to produce all individual 190 F1 cross combinations. The F1 seed was harvested, germinated and grown to maturity in a glasshouse. All ears were bagged to ensure self-pollination, and the F2 seed from each of the individual F1 plants of each cross was harvested and bulked for each cross. From these F2 bulks three separate ‘foundation’ composite cross populations (CCPs) were started by bulking F2 seed from the individual crosses. The first was synthesized from the 66 crosses between varieties with good milling potential (Q), the second synthesized from the 36 crosses for varieties identified as having high yield potential (Y) and the third synthesized from the 99 crosses between Y and Q parents (YQ). In addition, male sterile CCP populations (CCPms) were generated by artificially hybridizing all the above parents to characterized genetic males sterile lines (as females) obtained from two sources, RAGT and CIMMYT, shown below:

RAGT lines (Shango derivatives)CIMMYT lines

JB Plant 1 F1TOPDMSO102 7 TURACO DMS

JB Plant 2 F1TOPDMSO102 10 GALVEZ S 87 DMS JB NWH 65 F1TOPDMSO102 12 CUMPAS T88 DMS

F2/F3 Sterile Bulk Population 2/77 F1TOPDMSO102 14 NING8201 DMS

The F2s of these crosses were bulked as above to create QCCPms, YCCPms and YQCCPms, which together with QCCP, YCCP and YQCCP, gave six CCPs available as starting material for field evaluation.

Creation of mixtures

In order to compare the performanceof mixtures of homozygous lines with that of heterozygous populations, parental seed of equal proportions was also mixed in the same categories as those used to create the populationstoprovide a Yield, Quality and Yield-Quality Mixture. Unfortunately the Norman seed that was used in the mixtures subsequently turned out to have been contaminated. Therefore the Yield and the Yield-Quality Mixtures contained a small amount of an unknown genotype.

Seed bulking

The six CCPs, three mixtures and all parental varieties (apart from Norman) were drilled in single replicate plots of varying size at four locations in October 2003. There was enough seed available in autumn 2004 to begin replicated field trials.

Field trial site description

Trial sites consisted of two organic sites (Wakelyns Agroforestry (WAF), in Suffolk (52o39’N, 1o17’E); Sheepdrove Organic Farm in Berkshire (SOF) (51o41N, -1o52’E)) and two non-organic sites (Metfield Hall Farm (MET), continuous wheat, adjacent to WAF in Suffolk (52o41’N, 1o29’E); and Morley Farm (MOR), an experimental farm in Norfolk managed by The Arable Group (TAG) (52o56’N, 1o10’E)). Experiments took place in different fields within the rotations (or areas of fields) in each year on soils with clay content in the range 13-40 %. Soil type was medium to heavy at Metfield, Morley and Wakelyns and light to medium at Sheepdrove. The preceding crop was always grass-clover or grass-vetch-clover ley at Wakelyns; grass-clover at Sheepdrove; winter wheat at Metfield; and winter oil seed rape at Morley.

Assessment and analysis

Twenty two assessments were carried out each year on all plots at all sites throughout their growth cycle, from emergence to final yield and quality. CCPs and mixtures were analysed in relation to their contributing parents (Y, Q and YQ category). The Genstat package was used to analyse all data. Standard ANOVA and REML meta analysis was used to analyse individual year’s data. However, to analyse the reliability and stability of the populations and mixtures (for various attributes such as yield and quality), VNSI International developed a new method of stability analysis, ‘GEstability’, based on the cultivar-superiority measure of Lin & Binns (1988), which is explained below.

Results

Trial year one (2004/05) summary

The first year of replicated trials was completed in the cropping year 2004/05. In trial year one, the composite cross populations (CCPs) were at an early stage of environmental adaptation (see Phillips and Wolfe, 2005). The selected data presented demonstrates the variability across parent cultivars, mixtures and CCPs, and across field sites.

Grain yield

In both organic and non-organic systems there was a tendency for the mixtures and CCPs to produce a greater grain yield than the means of their parent cultivars (Figure 1a and 1b). This was more strongly evident at the organic sites.

a)

b)

Figures 1a & b. Mean grain yields for categories of Composite Cross Populations (CCPs), Composite Cross Populations with male sterility (CCPms), Mixtures (M) and Parents (P), all sites, in cropping years 2003/04 and 2004/05. Genotypic range is indicated forhigh yielding (Y), high quality (Q) or both high yielding and high quality (YQ).

For 2005, analysis of non-organic systems showed that the grain yield of YCCPms was significantly higher than QCCPms (p = 0.022) and in the organic systems, both the YCCP and the YQCCP out-yielded the QCCP (p = 0.013) (Table 1).

Grain yield (t/ha @ 15% mc)
CCP(ms) category / Organic system / Non-organic system
YQ / 6.4 b / 10.1
Q / 5.7 a / 9.8 b
Y / 6.3 b / 10.9 a

Table 1. Mean grain yield in non-organic (s.e.d. = 0.231) and organic (s.e.d. = 0.229) systems for categories Y (yield), Q (quality) and YQ (yield quality), 2004/05. Values indicated by the same letter in the same column are not significantly different (p < 0.05)

Comparing organic and non-organic sites, there was a clear difference in relative performance of the modern varieties, in that they produced high yields under non-organic conditions but relatively poor yields under organic conditions. Importantly, the populations exhibited a greater stability of yield across organic sites compared to their parents and, to their physical mixtures. However, this effect was not evident under non-organic conditions.

Trial year 2 (2005/06) summary

The replicated field trials of 2005/06included additional populations which had been exchanged among sites within systems. These multi-site populations provided a means of detecting any early site specific adaptation.

Grain yield

Although the CCPs performed within the ranges of the parents, they often yielded higher than the means of the parents. TheYCCPs had higher yields (P > 0.05) than the QCCPs at all sites. However, the YQ population yields did not differ significantly from the Y population yields at three of the four sites (MET, SOF & WAF), demonstrating an improvement in performance of the YQ relative to the Y populations since 2004/05.

At all sites except SOF, there were significant differences among the yields of the populations, mixtures and parental means; the mean of the parents was always significantly lower than the yields of thepopulations, and the mixtures. At both non-organic sites, the mixtures significantly out-yielded the populations, but this was not true at organic sites.

Quality indicators

There were highly significant (P < 0.001) differences in protein concentration among the varieties, mixtures and populations at all sites. When categories (Q, Y or YQ) were examined, significant differences were found at all sites; the Q category had the highest protein concentration and the Y category the lowest.For Hagberg Falling Number (HFN), there were significant (P < 0.005) differences between categories at two of the sites (MOR & SOF); the Y category had significantly lower HFNs than the other categories.

Trial year three (2006/07) summary

Populations and mixtures

The yields of the populations and the mixtures are compared to the relevant parent meansin Table 2 for the third season of replicated trials.

Table 2. Yields of the Y, Q and YQ populations, without or with male sterility, and the mixtures, as a percentage of the category parent means in 2006/07. Values of less than 3% above or below 100 are unlikely to be significant.

Non-Organic / Organic
Y / Q / YQ / Y / Q / YQ
Population / 103 / 103 / 101 / 102 / 103 / 109
Population with male sterility / 101 / 99 / 100 / 107 / 105 / 104
Mixture / 105 / 104 / 103 / 100 / 105 / 105

Although the yield gains from the populations and mixtures are relatively modest for the three years, they are consistent (see Table 2) with the larger gains tending to occur under organic conditions.

Integrated data from three years of trials

Organic conditions

All the populations consistently yielded more than the means of their parents and the YQ and Q mixtures out-yielded their parental means (Table 3). The YQ category was on average closer in yield to the Y than the Q category.

Non-organic conditions

The yield differences between populations,mixtures and parental means were smaller compared tothose in organic conditions, although the CCPs and mixtures did consistently out-yield the means of their parents. However, in the CCPms category only the YCCPms out-yielded the mean of its parents. The mixtures consistently yielded more than the populations. All the YQ categories were consistently intermediate between the Y and the Q categories (Table 3).

Table 3. Mean grain yield (t/ha @ 15% moisture content) for categories Y Q and YQ for three years at non-organic and organic sites.

Organic / Non-organic
Y / YQ / Q / Y / YQ / Q
Parents / 5.44 / 5.30 / 5.15 / 9.89 / 9.41 / 8.94
CCP / 5.57 / 5.62 / 5.18 / 10.17 / 9.47 / 9.17
CCPms / 5.75 / 5.44 / 5.32 / 10.01 / 9.39 / 8.84
Mixtures / 5.44 / 5.49 / 5.35 / 10.37 / 9.73 / 9.33

While the mean yields of the populations and mixtures are comparable to the mean parent yields, no indication of the stability of performance of these genotypes has been indicated. This aspect is considered in detail below (cultivar superiority).

Genotype by environment interactions

In agro-ecosystems the effect of the environment on crop performancecan be divided into the systems, sites and years. Inspection of the variances in the analyses of yield of all populations, mixtures and parents indicated a large component due to system but the effects of site and year were smaller.

Over three years, the yields of parents, mixtures and populations stayed at similar levels at MET although the mixtures tended to become more variable. In contrast, at MOR yields of populations and mixtures decreased to parental levels, although the populations provided the most reliable performance. At SOF there was little change over time except that mixtures tended to perform with greater stability over the three trial years. The populations were consistently better than mixtures for yield. At WAF the yields of populations and mixtures tended to increase with time.

Parent varieties

Varietal yields were closely correlated between the two non-organic sites, MET and MOR (r = 0.94, P<0.001). Correlation of yields between the two organic sites was also significant, but at a lower level (r = 0.60, P<0.01).

Stability

Under variable agronomic conditions, stability of yield (or other characters) over sites and years is of optimal importance to the farmer compared to mean yield alone. We therefore needed a measure that combined absolute performance with stability across environments. On the advice of VSNI International the cultivar-superiority measure of Lin & Binns (1988) was used for this purpose. For each genotype, this is the sum of the squares of the differences between its mean in each environment and the mean of the best genotype in that environment, divided by twice the number of environments. Genotypes with the smallest values have higher, more stable yields (or other measured character).

Cultivar superiority of combined factors

Cultivar superiority for all varieties, mixtures and populations was determined for seven developmental characters which contribute to the final performance of the crop (establishment, early crop cover, canopy cover, tillering, head density, harvest index and green leaf area) and five key characteristics important to the farmer (yield, protein, TGW, specific weight and total harvested biomass) here termed ‘harvest characteristics’. For each characteristic, the cultivar superiorities of the varieties, mixtures and populations were scaled to the appropriate parental mean, and then summed in the development and harvest categories. This provides a measure of cultivar superiority across a range of important characteristics and environments.

Confirming the central thesis of this project, both populations and mixtures proved their buffering capacity and displayed good cumulative cultivar superiority particularly for harvest characteristics relative to their pure line parents. For all characters in both systems, the mixtures and populations tended to have better than average cumulative cultivar superiority (Table 4 and 5). This was particularly notable for harvest relative to developmentalcharacteristics.

Harvest characteristics

In the YQ category (22 entries), the YQ mixture, YQCCP and YQCCPms had the top three cumulative cultivar superiorities in both organic and non-organic systems, andthe QCCP, QCCPms and Q mix performed best for cumulative cultivar superiority (out of 15) under organic conditions. In non-organic conditions the Q mix and QCCP performed better than all other genotypes for cumulative cultivar superiority, with the QCCPms in seventh place. For the Y category, the populations and mixture are in the top five genotypes, and under non-organic conditions, three of the best four genotypes for cumulative cultivar superiority (Table 4).

The mixtures were consistently lower ranking than the populations under organic conditions in all categories. However under non-organic conditions the mixtures were ranked first in the Q and Y categories, but in the YQ category the mixture was second to the YQCCP. The CCPms were lower ranking than the CCPs in both systems for all categories except in the Q category of organic systems.

Development characteristics

Performance of the populations and mixtures was more variable for development characteristics under both systems, although they were generally better than average (Table 4 and 5). The CCPs outperformed the CCPms in all categories under both systems, except in the Q category under non-organic conditions.

In contrast to the performance of the populations and mixtures, some varieties, although performing well in one system, performed considerably less well in the other. See for example in the Y category for harvest characteristics, Buchan (performing well under non-organic but poorly under organic conditions) and Wembley (where the reverse was true) (Table 4 and 5).

In relation to popular modern varieties, for example Claire, a commonly grown variety in organic systems, the populations and mixtures gave a superior performance in harvest characteristics. Similarly, Hereward a common variety grown under non-organic conditions was outperformed by the Q mix and QCCP.